It is been a while, I have been focused on an applied project in the field of real estate market.
I hope later I will write about this project, but for now following points describe the main elements of the project:
1- A mini-pc as a web crawler everyday collects all the listing advertisements in Switzerland and Germany. Currently we only crawl Homegate.ch in Switzerland and immobilienscout24.de in Germany , but other portals are on the way.
2- The advertisements will be projected back in a different medium in a form of spatial maps, where in addition to original adds, one can quickly see the average prices, rooms, sizes of listings per region.
3- Using these advertisements as some kind of spatio-temporal smaples, in parallel, we estimate the price index of all the buildings in Switzerland. This means 1.7 million buildings collected from OpenStreet Map.
4- While the previous item gives a rough estimate, there will be another form where a user can give specific attributes of his/her house, and the system estimates a range of possible rental values for that specific place. This part, which is not deployed yet will be using a sophisticated machine learning model to predict the rental/sale values for each house using the data from other advertisements and some other layers such as the distance to public transport or shops, etc. Since these estimated values are based on this goal to have a “Fair” price for a house, backed with market data, we call this process fair(y)+fication. Based on this we call this platform “FairyFy“.
We think there, we can define different kinds of business and end users to form a two sided market.
You can check the current working version of this portal here.